A protocol for data exploration to rapidly identify bias in at-sea observer or self-sampling program

To detect and subsequently minimise bias, this study investigates the utility of simple and universally-applicable quality indicators for two at-sea sampling techniques of Dutch demersal beam trawlers. We compared measures of proportionality of sampling on temporal, spatial and biological (i.e. catch composition) scales to evaluate how well observed trawls, trips and vessels represent their respective total populations. Between 2009 and 2011, vessels were sampled independently by using either observer-based or self-sampling techniques on different trips. Compared with self-sampling, observer sampling clustered in the Southern North Sea. During both observer- and self-sampled trips, sampling frequency was low at night, and on certain weekdays. Observers tended to overestimate total catch volumes, and seemed to underestimate numbers of benthic invertebrate discards. Using this protocol as a template to identify and reduce bias in at-sea monitoring schemes will be a useful tool to eventually improve the accuracy of observations.